ENERGY LOAD FORECASTING USING HYBRID K-MEANS CLUSTERING – SVM

Abstract

In Power Industry, Energy Load Forecasting is an important aspect. Determining the future demand for load in advance is very important. Once the company knows the future load, it can take much better investment decisions and decisions about expansion, maintenance and buying energy from the generating companies. Having some knowledge of future energy consumption is, therefore, an absolute necessity. Power distribution companies, therefore, require tools that can predict the load. Prediction of electrical load is difficult. A number of classical prediction models are available for this. But these models suffer from the problem of requirement of linearity and seasonality. For predicting electric load we have used K-Means Clustering and SVM. The results obtained using the technique are compared with energy load forecasting using SVM only and the performance of hybrid K-means clustering – SVM is found to be consistently better.